Dissemination and Implementation of Improving Pediatric Obesity Practice Using Prompts: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Dissemination and Implementation of Improving Pediatric Obesity Practice Using Prompts (NCT05627011), a randomized na study. The primary outcome is Mean Change in BMI as Percentage of 95th Percentile.

2. Clinical Question

3. Trial Design

This was a randomized, parallel, cluster-randomized trial of a behavioral implementation strategy. Practices, rather than individual patients, were assigned to iPOP-UP or usual care. This distinction matters statistically: patients treated within the same practice may have outcomes that are more alike than outcomes from patients in different practices. An analysis that treats every participant as fully independent could therefore misrepresent uncertainty. The registry identifies mixed models as the sponsor's analysis method, but does not report the model specification or the practice-level structure used in that analysis.

The intervention combined electronic health record clinical decision-support tools intended for pediatric overweight and obesity management. The control practices did not receive access to those tools, while retaining access to national opportunities related to the release of new American Academy of Pediatrics guidelines. The study therefore evaluated a practice-level implementation strategy in the setting described by the registry, rather than the isolated effect of a single clinical recommendation.

The registry reports enrollment of 58364 and lists the study as completed. It gives a start date of 2023-02-09 and a primary completion date of 2024-12-14. The listed phase is NA. The primary outcome timeframe extends up to 15 months after the first primary care visit following trial launch. The record does not state the number of practices randomized, the allocation ratio, or how many participants contributed data to the primary analysis. It also does not report whether allocation was stratified or constrained, or the procedures used to preserve allocation concealment. Masking is listed as none; additional details about who knew assignments during outcome measurement are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Only clinics were enrolled and randomized; no individual participants were directly consented/randomized. Of 92 clinics invited to participate in the trial, 81 clinics agreed/enrolled and were grouped into 71 clinic clusters and cluster-randomized to iPOP-UP vs usual care using covariate constrained randomization based on site, % patients publicly insured, % non-Hispanic White, geographic location (urban vs not), practice size, practice volume, and academic vs non-academic.

Overall Study

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Not collected for clinicians.

6. Primary Endpoint

The registry lists the primary outcome as Mean Change in BMI as Percentage of 95th Percentile, measured over Up to 15 months after first primary care visit following trial launch.

The primary endpoint is mean change in BMI expressed as a percentage of the 95th percentile. For each participant, BMI is related to the age- and sex-specific BMI at the 95th percentile using CDC growth curves. The registry states that a negative percentage indicates that the participant's BMI has decreased since trial launch. The measurement period is up to 15 months after the first primary care visit following trial launch.

This form of measurement places BMI in the context of an age- and sex-specific reference point, rather than interpreting the raw BMI value without that reference. In general, expressing a measure relative to a reference can make it easier to describe change across participants for whom the reference varies with characteristics such as age and sex. The endpoint is a continuous measure, and a mean summarizes the average change in the analyzed group. A mean does not show how outcomes were distributed among individuals or whether responses varied substantially.

The registry provides no observed means, mean changes, between-group estimate, or confidence interval for this endpoint. It therefore does not allow the magnitude or direction of the primary outcome difference to be assessed. The precise handling of repeated measurements, baseline values, incomplete follow-up, and any observations outside the stated timeframe is not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes the study as randomized, parallel, and cluster-randomized. It identifies the primary analysis method as mixed models. The available record is insufficient to reconstruct the model or reproduce its estimates, so the method can be explained only at a general level and in relation to the limited details that are reported.

What does cluster randomization change?

In a cluster-randomized design, groups such as practices are assigned to study conditions, while outcomes may be measured on individual participants. People treated in the same practice can share clinicians, routines, electronic records, and local implementation conditions. Their outcomes may therefore be correlated. In general, this correlation reduces the amount of independent information compared with an analysis that assumes every participant contributes entirely independent evidence. The number of practices and the degree of within-practice correlation are not reported in the registry record, so the effect of clustering on precision cannot be quantified from the posted information.

Cluster assignment also affects how intervention effects should be interpreted. The randomized strategy was assigned at practice level, and its estimated effect concerns that strategy in the participating practices and study context. It should not be treated as though each participant had independently been randomized to receive or not receive an electronic prompt. The registry does not report the number of practices assigned to each arm or the distribution of participants across practices.

What can be inferred from the mixed-model label?

In general, mixed models include terms that represent both average effects and sources of correlation or variation, such as clustering or repeated observations. Their structure can be adapted to different outcome types and data arrangements. The registry names mixed models for the reported analyses, but does not state which fixed effects or random effects were included, how time was represented, whether baseline outcome values were incorporated, or how standard errors and confidence intervals were calculated. It also does not describe any covariate adjustment. Those details are needed to assess the exact analysis and its assumptions.

The primary outcome is described as mean change in BMI as a percentage of the 95th percentile. The record does not specify whether the model analyzed individual changes, repeated outcome values, or another representation of the endpoint. Nor does it report whether the mean was adjusted for baseline characteristics. No model coefficients, estimated group means, contrasts, or interval limits are posted. Thus the reported p-values cannot be used to recover those quantities.

What does a confidence interval add?

In general, a confidence interval gives a range of values compatible with the data and the statistical model under the method used. Its width reflects uncertainty: a wider interval indicates less precision, while a narrower one indicates greater precision. The primary outcome identifies a 97.5% confidence-interval level, but no lower or upper limit is recorded. The interval's location and width cannot therefore be evaluated. A p-value does not substitute for those missing estimates.

How should the p-values be read?

The registry reports three superiority analyses, each with p < 0.001, using mixed models. In general, a p-value evaluates the data relative to a null hypothesis within a specified analysis. It is conditional on the model and its assumptions, and does not state the size of an effect. The registry does not give enough information to identify the contrasts represented by each group identifier. The p-values therefore cannot be confidently assigned to particular arm comparisons or interpreted as evidence favoring one named arm.

The label “superiority” describes the type of hypothesis test recorded; it does not supply the missing effect direction. To assess the result, readers would need the contrast definition, the estimated effect, and its confidence interval, along with information about the model. None of these are reported in the registry record.

Why is a simple recalculation not suitable?

A simple recalculation from arm-level summary statistics can sometimes check an unadjusted comparison. For a cluster-randomized trial, however, a calculation that treats participants as independently randomized can fail to account for within-practice correlation and yield misleading uncertainty. The registry explicitly indicates that such an individual-level re-analysis would be misleading and that only the sponsor's analysis is shown. In addition, the primary outcome's arm-level summary statistics are not reported. There is no independent recalculated estimate to compare with the sponsor's result.

What remains unknown about the analysis?

The registry does not report the sample-size justification, target effect, assumed clustering, planned interim analyses, or a detailed statistical analysis plan. It does not describe the analysis population, exclusions, missing-data strategy, handling of protocol deviations, or sensitivity analyses. These are not details that can be inferred from the study's enrollment or its completed status. They are important for understanding how the posted p-values relate to the planned analysis and to all randomized participants.

10. Analyses Reported by the Sponsor

The registry's inferential results are the sponsor's mixed-model analyses. It reports three superiority analyses, each with p < 0.001. No recalculated comparison from posted summary statistics is available. The record specifically notes that a simple individual-level re-analysis would be misleading for this cluster-randomized design, because it would not appropriately reflect the unit of randomization and the potential similarity of outcomes within practices.

The sponsor's analysis therefore takes precedence over an unsuitable individual-level recalculation; in this record, no such recalculation is presented. This does not mean the reported findings can be independently verified from the registry entry. The effect estimates, confidence-interval limits, model specification, and exact contrasts are not reported. The identifiers attached to the analyses are also not defined in a way that establishes which named arm comparison each p-value represents.

The available results support only a limited reading: the registry records statistically significant mixed-model superiority tests. They do not establish the size or direction of the difference, its precision, or its practical importance. Further interpretation would require the sponsor's complete estimates and analysis details, which are not reported in the registry record.

11. Multiplicity, Missing Data and Analysis Populations

The record lists 16 outcome measures and 58 analyses, including a primary outcome and secondary measures relating to clinician guideline adherence, diagnosis coding, screening, counseling documentation, follow-up requests, and referrals. Multiple analyses create a multiplicity concern: in general, when many hypotheses are tested, the chance of obtaining at least one small p-value by chance can be greater than for a single prespecified test. Whether the analyses were organized in a hierarchy, whether any adjustment was used, and which results were considered confirmatory are not reported in the registry record.

The primary outcome's timeframe is given as up to 15 months after the first primary care visit following trial launch. The registry does not report how many participants had outcome measurements at each point, how much data were missing, or how missing measurements were handled. It does not say whether the primary analysis used all randomized participants, a defined subset, or another analysis population. Reasons for missing data, withdrawal, or exclusion are not described.

In general, missing data can affect both the precision and the interpretation of an estimated treatment contrast, especially when the likelihood of missingness is related to participant characteristics or outcomes. Different assumptions and methods can produce different results. Without a stated missing-data approach or sensitivity analyses, it is not possible to assess how robust the reported analyses are to incomplete outcome information.

The registry does not report interim analyses, stopping rules, a detailed statistical analysis plan, or the sample-size justification. Their absence from the entry should not be taken as evidence that a particular procedure was or was not used; these details are simply not reported in the registry record. The enrollment total of 58364 does not by itself reveal the number included in the primary analysis or the information available for each outcome.

12. What the Result Does, and Does Not, Mean

The registry documents a randomized, cluster-level comparison of an electronic health record decision-support implementation strategy with usual care in pediatric primary care. It identifies a primary endpoint based on mean change in BMI relative to the 95th percentile and reports mixed-model superiority analyses with p < 0.001. These details describe the trial question and the inferential results recorded.

They do not show the estimated effect. No primary-outcome means, contrast, or confidence-interval limits are reported, so the direction and size of the difference cannot be read from the registry entry. The p-values do not tell whether the iPOP-UP arm or usual-care arm had a more favorable mean change. They also do not show whether any difference was large enough to matter in practice or how precisely it was estimated.

The registry's listed secondary outcomes concern care processes, including guideline adherence and documentation. Such measures address different questions from the BMI endpoint. The existence of those outcomes does not establish that any particular process changed, and their results are not included in the registry entries. The same caution applies to the number of outcomes and analyses: those counts describe the scope of evaluation, not the direction or success of any individual result.

In general, a trial's findings apply most directly to the study's participants, practices, intervention, comparator, and measurement period. The registry does not provide enough information about practice characteristics, participant characteristics, implementation fidelity, or analysis populations to judge how closely other settings match this study. The results should therefore not be extended beyond what the record reports.

Finally, the record's entries of zero participants at risk or affected for serious and other adverse events, and zero deaths, do not establish that no such events occurred in every relevant population or under all circumstances. Definitions, ascertainment procedures, exposure denominators, and safety follow-up details are not reported in the registry record. The entries should be read as registry fields, not as a comprehensive safety evaluation.

13. Why This Trial Matters Statistically

This study is statistically notable for assigning the implementation strategy at the practice level while measuring an outcome defined for participants. That design makes the unit of randomization central to analysis. In general, accounting for clustering is essential because participants within a practice may have correlated outcomes. The registry's use of mixed models and its warning against a simple individual-level re-analysis highlight why the design cannot be interpreted as an ordinary participant-level randomized comparison.

The primary endpoint also illustrates the importance of defining an outcome relative to an age- and sex-specific reference. Mean change in BMI as a percentage of the 95th percentile has a stated direction convention: a negative percentage indicates a decrease in participant BMI since trial launch. Understanding that definition is necessary before interpreting a reported estimate. Yet the registry provides no estimate or interval limits, leaving the central statistical quantities unavailable to readers.

The record also shows how an intervention may be assessed through both a participant-level health measure and a broad set of clinical-process outcomes. It lists 16 outcome measures and 58 analyses. This breadth makes transparent reporting of outcome hierarchy, multiplicity handling, and analysis populations especially important. Those details are not reported in the registry record, so the p-values cannot be situated within a complete analysis plan.

More broadly, a statistically small p-value is not a substitute for an effect estimate, confidence interval, or transparent description of the model. The study's posted results illustrate why those elements are needed to assess direction, magnitude, precision, and robustness—particularly in cluster-randomized implementation research.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The main interpretive limitation is that the registry reports p-values but no primary-outcome effect estimate, arm-specific means, or confidence-interval limits. The primary outcome includes a 97.5% confidence-interval label, but the interval itself is not reported. As a result, neither the direction nor magnitude of the primary comparison can be assessed from the posted results.

The three reported analyses are identified by group codes whose meanings are not explained in the record. Although each is labeled a mixed-model superiority analysis, the exact contrasts are unclear. The model structure, covariate adjustment, treatment of repeated observations, and handling of practice-level clustering are also not reported. The method name alone does not permit evaluation of model assumptions or reproduction of the analysis.

Important design and data details are absent, including the number of practices, the allocation ratio, practice and participant characteristics, the primary analysis population, and the number of participants contributing outcome data. The record does not describe missing-data methods, sensitivity analyses, implementation fidelity, or how the intervention was used in routine practice. These omissions limit interpretation of both the estimated effect and its applicability to other settings.

The record lists 16 outcome measures and 58 analyses, but does not report a statistical analysis plan or whether multiplicity was addressed. It also does not state the sample-size justification or whether interim analyses were conducted. The adverse-event entries contain zeros, but event definitions, ascertainment, follow-up, and safety denominators are not reported. These limitations reflect missing details in the registry record and should not be resolved by assumption.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Dissemination and Implementation of Improving Pediatric Obesity Practice Using Prompts — NCT05627011.
• Registry record last updated 2026-03-19; status last verified 2026-03.
• Results first posted 2026-03-19.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT05627011 is a completed, randomized, parallel, cluster-randomized study sponsored by Yale University. It evaluated iPOP-UP, an electronic health record-based clinical decision-support strategy for pediatric overweight and obesity management, against usual care. The registry reports enrollment of 58364 and lists a start date of 2023-02-09 and primary completion date of 2024-12-14.

The primary outcome was mean change in BMI as a percentage of the 95th percentile, assessed up to 15 months after the first primary care visit following trial launch. The registry records three mixed-model superiority analyses, each with p < 0.001. It does not report effect estimates, interval limits, or the meaning of the analysis group identifiers. Consequently, the direction, magnitude, and precision of the primary comparison cannot be determined from the registry record.